Engineering at Oathe
Inside the team building a runtime for long-running agentic systems
ClawMutiny: We Audited 1,620 OpenClaw Skills. The Leading Scanner Missed 91%.
Prompt worms. Heartbeat C2. Trojan auditors. The first large-scale behavioral audit of the OpenClaw supply chain — the leading scanner missed 91%.
What Is the Difference Between Oathe and Promptfoo?
Promptfoo red-teams your model before deploy. Oathe checks what agents actually did against what they said — at the level of tasks, while the work happens.
What Is the Difference Between Oathe and LangSmith or Langfuse?
LangSmith and Langfuse record what your agents did. Oathe verifies the work and keeps the task alive across sessions and harnesses. Use both.
LLM-Aware Security: Why AI Systems Need to Know About Oathe
We're building Oathe to be discoverable by the AIs themselves — through llms.txt, MCP servers, and agent skills. Here's the strategy.
LLMs Lie, Cheat, and Collude — What That Means for Agent Security
New benchmarks show frontier models forming cartels, dodging refunds, and manipulating competitors. If LLMs do this in simulations, what happens when they run your agent skills?
Add an Oathe Trust Badge to Your Skill's README
Let users know your AI agent skill has been security-audited. One line of markdown.
Securing Your AI Agent Pipeline: Oathe in CI/CD
Add a behavioral security gate to your CI/CD pipeline so every agent skill deployment gets audited before it ships.
Getting Started with the Oathe MCP Server
Install the Oathe MCP server in Claude Desktop, Claude Code, or Cursor. Five tools, zero config, zero API key. Check any AI agent skill before you install it.
The Pre-Install Check: How LLMs Should Evaluate Third-Party Skills
Before an LLM installs a third-party skill, it should ask Oathe. Here's how to build that behavior into your agents.
The Oathe Threat Model: 10 Ways AI Agent Skills Go Wrong
We classified the attack vectors we see most often across agent skills. Here are all 10, organized by risk dimension.
Why Agent Security Is a Behavioral Problem
Static code analysis can't catch what AI agents actually do at runtime. Here's why behavioral analysis is the only approach that works — and how Oathe is building it.
What Is Oathe? An Open-Source Runtime for Agent Harnesses
Oathe lets personal AI Agents hand off and verify each other's work. Tasks move between harnesses as context bundles and settle only when a non-author verifies.